I heard from a DFL source recently that House Speaker Melissa Hortman has sworn off polling.

Not true, the Brooklyn Park Democrat told me. But not out of the ballpark, either.

"I don't love polling, no," Hortman said. "I don't like spending money on things that don't provide value."

Everyone, it seems, is down on polls.

A common refrain is that President Donald Trump defied the polls. It's really only half right. On the eve of the 2016 election, national polls were close to predicting his 2 percentage-point loss in the popular vote.

Battleground state polls were wrong, however, and they were all wrong in the same direction.

Educational attainment has become a key dividing line between the parties. So a miscalculation about the educational composition of the electorate was bound to distort the picture, and that's what happened in key Midwest battleground states.

But that's just the start of the problems, Carleton College political science emeritus Prof. Steven Schier told me in e-mails and a phone interview.

Here are a few:

If you ask a one-sided question, it may turn into what's called a "push" question, which is lingo for a biased inquiry that is seeking to drive public opinion, not measure it. So for instance, "Do you believe the president is guilty of impeachable offenses?" You've already introduced his guilt into the question.

Here's an alternative: Which do you agree with: The president is guilty of impeachable offenses or this an unfair partisan attack?

These two-sided questions are longer and more difficult to write, which means more expensive.

Questions about race can be problematic. Schier said political science has shown that the perceived race of the questioner can affect the response on race issues.

Then there's the cellphone problem: What percentage of respondents should be on cellphones? Polling cellphones is much more expensive.

And what portion of respondents comes from communities where English is a second language?

Polls also are not effective at measuring the public's intensity on an issue.

A respondent might say she favors more gun control, but that doesn't measure her willingness to go out on a snowy day and vote on that issue — and that issue alone.

"Intensity drives politics," Schier said.

He cites the work of political scientist David RePass, who has shown that by asking the same set of respondents the same questions over a period of months, we find very few people have consistent views.

"It means a lot of people aren't thinking much about the issues," Schier said.

The public is also prone to wide swings in the face of major news events.

Yes, the media are partly to blame. But Schier said the public is also responsible: "The voice of the public is not always the voice of God. God is a little more alert."

J. Patrick Coolican 651-925-5042 • @jpcoolican patrick.coolican@startribune.com